1. INTRODUCTION.
This "page" looks at some of the problems the FAO Statistics Division (ESS) faces in compiling agricultural statistics; it doesn't attempt to go into the problems facing the countries themselves - which are probably well known; it is not a guide to data collection in countries. Therefore, we will not discuss the statistical organization of the countries, their sources of information, methods used by them for collecting data and concepts and definitions adopted. At the outset we do note that sources and methods which are very useful in a one country, perhaps are not at all applicable in another country. The message that we try to disseminate across the Internet is that countries should strive to collect and report their figures in conformity with the guidelines and recommendations given by FAO, as regards concepts, definitions and coverage of the data.

2. THE LARGE VOLUME OF DATA.
A leading difficulty faced by ESS arises from the enormous data that are required to be manipulated, and the multiplicity of the sources reporting the data.There is a huge quantity of annual figures relating to various elements of the supply utilization accounts (SUAs) to be collected or estimated and then evaluated and recorded. Two examples: just the element in the accounts (031)- area or input - includes about 25000 country entries for a given year; some 5500 of them relate to harvested area of crops, while most of the other 20000 relate to "input", "slaughterings" and "animals producing".

The magnitude of the figures - in terms of units - moves between a few thousands hectares (HA) harvested of various crops in many countries and 42 million HA of rice harvested in India.

Element 051, Production/Output, includes more than 28500 country entries per year, of which about 8000 refer to production of primary crops and production of those few crop processed products shown in the FAO Production Yearbook. Production figures moves between a few thousand tons of various crops produced in many countries and more than millions tons produced in some countries; for example, 280 million tons of sugar cane in Brazil and 250 million in India; 255 million tons of maize in the United States; 190 million tons of paddy in China; 150 million tons of maize for silage in the Russian Federation.

3. DATA SOURCES FOR ESS.
a) Primarily, FAO Questionnaires. These are tailored questionnaires and sent to member countries of FAO.
b) National publications available in the ESS Library. There are some 3700 titles in all, including general yearbooks, agricultural yearbooks, pocket yearbooks; periodicals (quarterly, monthly, weekly, daily, etc.); early warning system correspondents' reports; FAO Country Representatives' reports.
c) Country visits by ESS statisticians involving discussions with national experts.
d) International publications,

• - Oil World bulletins for oil crops and derived products; butter and animal fats. For important countries, also production and trade of meat and feed, livestock numbers, hatcheries' statistics.
• - Licht's reports on sugar, molasses, sweeteners and as a second source, coffee.
• - EEC Agricultural Service International.
• - Agri-Afrique.
• - Bulletin de l'Afrique Noire.
• - Cotton Advisory Committee bulletins, quarterly and monthly.
• - Bulletins DPA: (Poultry, Fruits, Vegetables, Wine, Milk, Meat, Potatoes).
• - International Egg Commission bulletins.
• - International Tea Commission bulletins.
• - International Rubber Study Group bulletins.
• - ZMP balances (mainly for Germany and EEC): Milk; Meat, poultry, eggs; Potatoes; Fruits; Vegetables.
• - International Cocoa Organization bulletin.
• - International Coffee Organization bulletin.
• - Marchés tropicaux et méditerranéens.- REUTER telex.
• - BBC printed daily and weekly reports.
• - News Bulletin of Xinhua News Agency.
• - The Public Ledger.
• - The Financial Times.

In some cases, official figures released by the countries can be found in these publications long before they reach our office through questionnaires or national publications.

It is not always easy to ascertain which one of the various figures reported by the various sources is the most recent or the most reliable, although the national Yearbooks data are supposed to be "the best ones". However, they report historical data for a number of years with the last year shown being one to three years behind the current year.

It is always necessary to evaluate the data for consistency within the country and between countries, and to make sure that definitions and coverage of the data conform with FAO recommendations. To do that, it is indispensable that those in charge of collecting data have a fairly good knowledge of the various commodities they handle every day, from the point of view of agro-economic and biological characteristics. For example, they must know the difference between flax retted and flax scutched and hackled, between sown and harvested area, between dextrose and isoglucose, fruit juice and fruit nectar, carcass weigth, dressed carcass weight and ready-to-cook weight, etc. They should have a good idea of the possible and actual yield per hectare of various crops in different countries, take-off rate of livestock, potential and actual yield of milking cows, laying hens, etc.; the average weight of edible offals and slaughter fats of various species of animals, etc. Short descriptions of the commodities/elements can be found in the "Introduction", "Explanatory Notes" and "Notes on the tables" of the FAO Production Yearbook. Also, in the recently issued booklet "Definition and classification of commodities".Other related pages on this site are alos very useful in this regard.

4. PROPER USE OF DATA SOURCES.
When analyzing the data from sources, different from " the main source " or the traditional source used, one important task is to carefully check whether the data are comparable or not with the series shown , from the point of view of coverage, definitions, time-reference, etc. You can't just plug in data like an internet's "plug 'n play" tool. For that purpose, it is important to take note not only of the data for the current year, but also for previous years. Even if the data are not comparable, they can still be of some use, at least as year to year variations in percent terms. Moreover, for a better comparison of the data given by different sources, it is important to note the date when such figures have been released rather than the date of the publication reporting them. Also important is to note from which office they emanate: Statistical Office, Ministry of Agriculture, Planning Boards, commercial firms, etc., etc.

Because of human error, it can happen that official data are clearly wrong and inconsistent with other available information and common sense, e.g., the wheat yield may be reported as 200 tons per hectare. In those cases, official data are to be replaced by estimates, after having tried to clear the matter with the countries concerned.

On the possibility of utilizing Census data for adjustment of current agricultural statistics, it should be observed that this requires careful ascertaining the quality of the census results, to make sure that they were more reliable than "current sample statistics". However, we know by experience, that not always Census data are more reliable than current data, particularly as regards production and coverage of area and production figures.

5. MAKING ESTIMATES OF PRODUCTION.
The following are the two most important elements to be taken into account in making estimates of production:
- For crops: sown and harvested area and yield per hectare.
- For livestock: number of producing animals and yield per animal.

In many developed countries and in a few developing countries, changes in the area sown are primarily due to changes of government policies, inter alia. In fact, area changes in accordance with what the governments want to produce. To this aim, governments fix intervention prices, move up and down guaranteed prices and production ceilings, change production subsidies and put into operation various types of set-aside area programmes. In developing countries, changes in sown area are due principally to the weather and soil conditions at the time of sowing and also to the availability of seed. Needless to say that expectations of changes in national and international prices are also taken into account by farmers in deciding the level of the sowings. For example, one year could be economically more convenient sowing more cotton and less tobacco.

Changes in the yield per hectare are primarily due to weather conditions during the growing period, secondly to the more/less, better/worse, use of inputs, such as fertilizers, pesticides, farm machinery, irrigation, agro-technology in general. Yields per animal depend primarily on the amounts and quality of feed, apart, naturally, from the aptitude of the various races to produce meat rather than milk or eggs or wool or vice-versa. Yields per hectare or per animal, as well as take-off rates, can be seen country by country in the FAO Production Yearbook, or can be easily derived from there. Naturally, there are enormous variations from country to country. For example, maize yields go from 0.5 to 10 tons per hectare; milk per cow, from 200 to 7 000 kgs.; take-off rate of cattle from 6 to 50 percent, wool per sheep from less than 1 kg., to 6 kgs.; cotton lint/seed cotton ratio moves from 26 to 40 percent.

Annual production estimates of meat, milk, dairy products and eggs can be based on monthly production figures or deliveries to dairy plants released by most developed countries. Also estimates of certain crops and crop products are based on monthly figures, for example, tea, sugar, cocoa, rubber and palm oil.

Monthly imports/exports figures are also used to estimate annual feed supplies for important commodities, such as cereals, bran, oilseed cakes, cassava dry and maize gluten.

Production estimates of a primary commodity can be made on the basis of utilization data, when available, e.g., exports, food, etc.; and also on the basis of production of processed products derived from the commodity for which data are missing.

Since indigenous meat production is defined as "slaughtered meat production plus the meat equivalent of animal exported alive minus the meat equivalent of animals imported alive", it follows that monthly import/export figures of live animals are also necessary to estimate indigenous meat production.

In this connection, since import/export figures of live animals are generally given by countries in numbers rather than weight, it is important to note what kind of animals (big or small) are imported and exported. For example, the meat equivalent of 2 million day-old-chicks is about 80 tons while the meat equivalent of 2 million chickens and hens is 2 200 to 3 000 tons. Meat equivalent of 2 million calves is 70 to 250 thousant tons; 2 million adult cattle, 320 to 600 thousand tons; 2 million lambs, 20 to 30 thousand tons; 2 million sheep, 30 to 50 thousand tons.

So, the "average carcass weight" of bovine animals imported or exported, half calves and half adult, will move between 100 and 210 kg/head.

6. ESTIMATION OF MISSING DATA.
Although there are so many sources of data, it should be noted that most sources deal with cash crops or with crops and livestock production which can influence the international markets. Production of crops and livestock, whose main or exclusive use is the subsistence of producers, is seldom covered by these sources; in other worlds, too much information for certain commodities and too little for others. Result, missing data and gaps in the series.

In the absence of any figure from any official or unofficial source, estimates for missing data of area and production of a commodity can be made on the basis of various assumptions, the safest one appears to be following the observed trend of the commodity in question in the previous years, with an eye also on the behaviour of that commodity in neighbouring countries.

The estimation of a complete time series, when neither official nor unofficial data are available, presents more difficulties. When we know from various sources, e.g., Trade Reports, Food Balance Sheets, Households Surveys, Census reports, etc., that a certain crop or a certain livestock product is produced in a given country, some (rough) estimates can be worked out on the basis of the little information available, and taking into consideration the situation of this commodity in neighbouring countries with similar agro-economic conditions and similar food habits. Contacts should be established with the countries concerned, directly or through FAO or UN Officers living or visiting the countries, so that at least some first-hand information is made available to ESS. All this, of course, when the matter deserves such an attention.

Conclusion.
After having examined this page we hope you have gained a small appreciation of the complexity involved in (The Statistics Division) preparing data for users such as yourselves and it is hoped that the statistician's work obtains the recognition it rightly deserves. It should be clear that our search in compiling data is for "the truth" and the statistical true estimate and that requires your assistance. We invite you to use the Data Contributions Form found on our web page and be part of this deserving quest. Thank you.